cancel
Showing results for 
Search instead for 
Did you mean: 
Highlighted

only ndmp path goes down

We have 10 drives configured from Quanutm i500. All drive are shared with 6 windows media servers. Master is VM and does not control robot.

Among these, 4 Drives are also configured (shared) with 2 Netapp hosts ( 2 each). NMDP drives are configured to media server bck04 which is also the robot control host.

NetApp ontap - 8.1.2 (7-Mode), NBU 7.5.0.6 on windows 2008 R2 64 bit.

The problem is after the nmdp backup is successfully done the netbackup tries to umount  the tape but it fails and eventually down only the NDMP path.

ltid logs

13:04:55.396 [4920.3400] <4> tpunmount_withMntPtr: is_ndmp
13:04:55.396 [4920.3400] <4> tpunmount_withMntPtr: drive has been assigned
13:04:55.396 [4920.3400] <4> add_unload_to_queue: Unload for media id D00270 added to queue (PID 2892, OP 2)
13:04:55.396 [4920.3400] <2> SendRobotReq: LTID - Sent ROBOTIC request, Type=3, Param2=6
13:28:58.415 [4920.3400] <6> WriteEntry: Updating drive QuantumDrive006 at path nrst0a on attach host fas3240x01a
13:28:58.525 [4920.3400] <4> LtidProcCmd: Sending UNLOAD DONE msg to BPTM (dev 6, PID 2892)
13:28:58.525 [4920.3400] <4> send_request_answer: Pid=2892, Type=330, ReturnType=12, DataType=2, MsgRes.SC=10
13:29:05.623 [4920.3400] <6> DownDrivePath: DOWNing path nrst0a for QuantumDrive006 on host cmsbck04.mycmsc.com
13:29:05.623 [4920.3400] <6> WriteEntry: Updating drive QuantumDrive006 at path nrst0a on attach host fas3240x01a

I have attached the robot logs.

The backup runs fine on other drives and netbackup does not down other paths. We have to up the ndmp path everyday so that ndmp backup completes.

1 Solution

Accepted Solutions
Highlighted
Accepted Solution!

I have had some strange

I have had some strange device issues with NDMP. I think what happens is if a NetApp happens to reboot during a backup, the NetApp device tree gets messed up and you can end up with too many device aliases. My process to alleviate this is:

  1. On the filer, run "ndmpd killall" to kill any outstanding NDMP sessions.
  2. On the filer, run "sysconfig -t" to see if there are any tape drives defined (there should be.)
  3. On the filer, remove all tape drive aliases "storage unalias -t"
  4. Determine which adapter on the filer is connected to the tape drives. The method I use is to run a "sysconfig -a" and find which adapter has tape drives defined - not disk drives. Most of our filers seem to use adapter 1, so the adapter port could be 1a, 1b, 1c, etc.
  5. Reset the adapter on the filer. BE CAREFUL! You do not want to reset a disk adapter! This will also throw some errors that will trigger some SNMP traps. Here's an example of what we do. Warning! Make sure you are using the appropriate adapter port for your filer!!! This is just an example!
    "storage disable adapter 1c"
    "storage enable adapter 1c"
  6. Re-create the tape aliases "storage alias"
  7. With NetBackup, make sure no backups are running to the library/tape drives, you will be re-scanning.
  8. Delete the NDMP drives from the NetBackup configuration. The device manager services will have to be restarted (on the server controlling the robot.)
  9. Rescan for the NDMP tape drives - I usually use the device wizard. I only specify the hosts I need. Make sure the NDMP box is checked and your filer names are entered correctly. Then scan for the devices and restart the device manager service. 
  10. I don't remove any storage units, so I don't let the wizard create any storage units.

I don't know if this solves your problem or not. But when my devices get confused, this fixes it.

 

View solution in original post

4 Replies
Highlighted

Been a while since I worked

Been a while since I worked with NetApp, but then there waas an option for SCSI Reserve that should be set to the same as Netbackup uses.

The standard questions: Have you checked: 1) What has changed. 2) The manual 3) If there are any tech notes or VOX posts regarding the issue
Highlighted

Its same on both ends. SCSI

Its same on both ends. SCSI Reservation set on both NetApp and NetBackup

Highlighted
Accepted Solution!

I have had some strange

I have had some strange device issues with NDMP. I think what happens is if a NetApp happens to reboot during a backup, the NetApp device tree gets messed up and you can end up with too many device aliases. My process to alleviate this is:

  1. On the filer, run "ndmpd killall" to kill any outstanding NDMP sessions.
  2. On the filer, run "sysconfig -t" to see if there are any tape drives defined (there should be.)
  3. On the filer, remove all tape drive aliases "storage unalias -t"
  4. Determine which adapter on the filer is connected to the tape drives. The method I use is to run a "sysconfig -a" and find which adapter has tape drives defined - not disk drives. Most of our filers seem to use adapter 1, so the adapter port could be 1a, 1b, 1c, etc.
  5. Reset the adapter on the filer. BE CAREFUL! You do not want to reset a disk adapter! This will also throw some errors that will trigger some SNMP traps. Here's an example of what we do. Warning! Make sure you are using the appropriate adapter port for your filer!!! This is just an example!
    "storage disable adapter 1c"
    "storage enable adapter 1c"
  6. Re-create the tape aliases "storage alias"
  7. With NetBackup, make sure no backups are running to the library/tape drives, you will be re-scanning.
  8. Delete the NDMP drives from the NetBackup configuration. The device manager services will have to be restarted (on the server controlling the robot.)
  9. Rescan for the NDMP tape drives - I usually use the device wizard. I only specify the hosts I need. Make sure the NDMP box is checked and your filer names are entered correctly. Then scan for the devices and restart the device manager service. 
  10. I don't remove any storage units, so I don't let the wizard create any storage units.

I don't know if this solves your problem or not. But when my devices get confused, this fixes it.

 

View solution in original post

Highlighted

Thanks miller. I will work on

Thanks miller. I will work on this and share the udpate here.

Other Experts can share their thoughts and input & I will check how can I reverify from my end..